Storm of Money: Hurricanes and Insurance
Storm Of Money: Hurricanes, Insurance, And The Secret Black Boxes That Set Our Rates
June 05, 2012
By TONY BARTELME
The Post and Courier
Some things are certain: As the earth spins, air moves swiftly around the equator, creating the trade winds.
It's also certain that storms will form because the sun shines bright where these trades blow, turning sea water into sky-high clouds of steam that inevitably collapse, a process announced by torrents of rain and thunder.
And we know for sure from history and physics that a few of these air masses will spin counterclockwise, slowly at first, then faster and with enough momentum to flatten cities, alter destinies, and if hooked into some fantastic electric grid, pack enough energy to light every bulb on earth.
Beyond these certainties, hurricanes challenge us with their unknowns. They are so infinitely changeable that it may be mathematically impossible to predict their exact paths more than a few weeks in advance.
Because we can't predict the future, the beginning of hurricane season is a time of warnings and pleas to prepare. Like the humidity, anxiety settles in for the summer.
South Carolinians also try to tame this uncertainty by forking over $1.3 billion in homeowner's insurance premiums every year. How insurance companies set these rates is a story that touches all of us, but especially those who own homes in coastal counties.
It's a story that leads to globally recognized scientists, including some with data that show a home on the coast might not experience catastrophic hurricane winds for hundreds of years. It features profit-seeking insurance companies that hiked rates, frustrated homeowners who scrambled to pay for these increases, and brilliant researchers who invented controversial computer models dubbed "black boxes."
It shines a light on South Carolina's insurance regulators, who for the past decade watched rates soar but did almost nothing to find out how these secret black boxes truly affect homeowners' rates.
It could begin almost anywhere, but why not start at a cocktail party?
Fear of hurricanes
Three years ago, Daryl Ferguson and his wife were mingling with friends in Beaufort when a familiar topic came up: "Have you been watching TV?" a retiree from Connecticut asked Ferguson. "A hurricane is tracking right at Beaufort County."
Another friend chimed in: "They don't hit us that often, but when they do it's awful. We almost got wiped out in 1893." These fears got personal when two friends told Ferguson after church that they were moving back to Ohio. "I said, 'You guys love the Lowcountry, why are you leaving?' And my friend said that his wife just worries about the hurricanes all the time."
Ferguson began to wonder: What is the true risk?
Ferguson, 73, isn't your ordinary retiree. He jabs the air when he makes a point and has an authoritative voice that seems to echo even in a carpeted room. For 10 years, he was president and chief executive officer of Citizens Utilities, then the largest diversified utility company in the nation. Between 2000 until his retirement in 2005, he was chairman of Hungarian Telephone, sometimes matching wits with Russian mafia figures. "I like complex problems," he said. So three years ago he began to learn about the complexities of hurricanes.
Ferguson's first call was to the National Hurricane Center, the National Oceanic and Atmospheric Administration's concrete fortress in Miami. He spoke to the head of the center's science unit and other meteorologists. He filled binders with charts and notes.
Over time, Ferguson learned how the jet stream undulates like a dropped firehose across the United States, but that it tends to twist south in the fall, sending tropical storms south into the Gulf of Mexico or north toward the Outer Banks, away from South Carolina.
He felt more reassured as he discovered that wind fields on the west side of hurricanes usually are smaller and weaker than on the east side. He learned about a quirk in South Carolina's geography.
South Carolina is triangular, like a poorly cut piece of pie, with the crust side facing the ocean. This triangle is positioned in such a way that the northern part of the coast juts about 150 miles farther into the ocean than the southern section.
Put another way, North Myrtle Beach is 150 miles closer to the Atlantic hurricane lanes than Hilton Head.
The concave shape of the coast makes the area from Charleston to Jacksonville less vulnerable to hurricanes, said Chris Landsea, science and operations officer at the hurricane center, one of the experts Ferguson contacted. "South Carolina does indeed get struck by major hurricanes," Landsea said, but the state's risk "is relatively lower" than Florida, coastal Louisiana and North Carolina'sOuter Banks.
Ferguson heard similar statements from other forecasters. "I was stunned. They all kept saying the words 'relatively low risk.'"
Relatively low risk? In Ferguson's mind, that went against the conventional wisdom that South Carolina is a hurricane magnet. It also seemed at odds with history.
Since 1851, 30 hurricanes had spun within 50 miles of South Carolina, according to NOAA records. That was one about every five years, which seemed to be the definition of vulnerable. But a closer look at the data was revealing: Of these 30 storms, 23 were minor hurricanes, such as 2004's Charley and Gaston, which caused minimal damage. The remaining seven had winds greater than 110 mph and indeed wrought devastation where they had gone ashore. The Sea Islands Hurricane of 1893 killed as many as 2,000 people around Beaufort; Hugo in 1989 took 26 lives in South Carolina and caused $6 billion in damage.
But even these catastrophic hurricanes often left large swaths of South Carolina unscathed. Hugo, for instance, did little damage south of Seabrook Island, and more than half of the state saw winds of less than 60 mph.
So while it was correct to say seven major hurricanes had touched some area in South Carolina since 1851, about one every 23 years, that didn't really say much about the long-term vulnerability of a particular house in Beaufort or Charleston. And in the end, Ferguson asked himself: Don't people want to know what's likely to affect them personally?
When scientists look at specific locations instead of areas as large as states, the odds of a home being hit by a hurricane change dramatically.
Kerry Emanuel is a professor of atmospheric sciences at the Massachusetts Institute of Technology and one of the world's leading authorities on hurricanes. At The Post and Courier's request, Emanuel's company, WindRisk Tech LLC, looked at 14 places along South Carolina's coast to see what kinds of winds these spots might experience over time.
Most scientists use historical data on wind speeds, barometric pressures and other variables to make their predictions, even though they acknowledge that this data has its shortcomings. Some observations go back to the mid-1800s, but truly reliable measurements have been taken only in recent decades.
Statistical analyses typically need vast amounts of data to be accurate, but Emanuel and other modelers overcome this dearth with a neat bit of mathematical sleight of hand: Using principles of physics and other factors, they generate tens of thousands of virtual hurricanes on their computers. Doing this creates an archive equivalent to 5,000 years of storms, he said.
The computer simulations showed that a point in Charleston was likely to experience 74 mph winds, or a minimal Category 1 hurricane, about every 37 years over a 5,000-year period. The model showed that a catastrophic hurricane, one with 115 mph winds, would on average affect a spot in Charleston once every 370 years. The situation on Hilton Head was even less threatening. A Hilton Head homeowner would be expected to experience a minimal Category 1 storm every 51 years, and a Category 3 storm about every 430 years, the WindRisk model showed.
Emanuel, Landsea and other experts cautioned that these calculations, known as "return periods," don't predict the actual timing of a hurricane's next visit.
"Hurricanes don't care what happened last year," Landsea said. "You can get hit twice in one year, or one year after the next. So the return period is more of an abstraction." But the studies do give people some idea of their vulnerability to hurricanes.
To Ferguson, data like this was a bombshell.
It meant that while South Carolina as a whole was likely to experience minimal hurricane-force winds every few years, the risk of those winds affecting him personally were about once every half century.
It meant that the truly disastrous ones were rare and certainly not the impending train wrecks that weather channels and emergency planners sometimes suggest are on the way.
But if that was all true, he wondered, why were his insurance rates so high?
What we pay
South Carolinians pay on average about one-third more for their homeowners' insurance than property owners in North Carolina and Georgia, according to a Post and Courier analysis of data from the National Association of Insurance Commissioners. Last year, South Carolina insurers and the state's wind pool collected $1.3 billion in home insurance premiums, nearly three times what they charged in 1996. Statewide, average premiums have risen 71 percent during the past decade.
But this average understates the impact in coastal counties. The average premium for someone with $150,000 in insurance is about $2,000 in Charleston County and $1,840 in Beaufort County. The farther inland you go, the less you pay. In Berkeley County, the average premium is about $1,200; in Dorchester County, it's $1,000, and upstate in Greenville County, it's only $720, according to S.C. Department of Insurance records.
Today, it's not unusual for some Lowcountry homeowners to pay more in insurance than property taxes. How did it get so bad?
The S.C. Insurance News Service, a nonprofit group funded by insurance companies, cites a mix of factors: dramatic growth in coastal South Carolina; rising property values; increased building costs; and new meteorological predictions that the world has entered a period of higher storm frequencies.
But it's also helpful to rewind to 1989, when Hugo rammed like a cannonball into South Carolina's midsection. John Richards was the state's insurance commissioner at the time. "We hadn't had much experience with catastrophes in the modern era until Hugo."
The storm caused $6 billion in damages, which at that time made it the most expensive insurance disaster in the nation's history.
It also was a shining moment for the industry. Teams of agents swooped in. Amid the debris, they often wrote checks on the spot; billions of insurance dollars spurred a building boom. Richards said only two small companies were shut down because they were insolvent. Lawsuits were rare. "The insurance companies did a wonderful job after Hugo," he said.
But it marked the end of an era -- in meteorological terms and in insurance company boardrooms.
Scientists had long wondered why there had been a lull in Atlantic hurricanes from the 1940s until the early 1990s. Some researchers theorized that it was a natural climate cycle called Atlantic Multidecadal Oscillation. Then Kerry Emanuel and Michael E. Mann, a prominent Penn State climatologist, noticed something else: Hurricane frequency seemed to correspond to pollution levels in the tropics.
They found that pollutants rose dramatically from the 1940s through the 1970s as the world's industrial production grew. These pollutants reflected sunlight, reducing the amount of energy hitting ocean waters in the tropics, the heat source that gives birth to the storms.
"We had suppressed them," Emanuel said of the hurricanes.
But those pollutants began to dissipate with new clean-air laws in the 1970s and 1980s. More sunlight reached the tropics by the early 1990s. Emanuel and Mann theorized that the warming oceans in the tropics would fuel a frenzy of new storms, a theory supported by subsequent studies.
Meanwhile, insurance companies thought Hugo was an aberration, said Richards, the former South Carolina insurance commissioner. "They thought, Thank goodness, this is just a storm we'll never see again for a decade or two."
Even after Hugo, insurance experts assumed that storms in heavily populated Florida would cause damages in the low billions of dollars, losses they figured they could swallow.
Then Andrew hit South Florida in 1992. It was a compact hurricane, more like a giant tornado; hurricane-force winds formed a bullet of wind 40 miles wide that entered south of Miami, churned through the Everglades and exited into the Gulf just four hours later.
Inside this maelstrom, 140 mph winds hammered the city of Homestead and other communities. Richards rushed south to help his colleagues. "The night I arrived, I was in Tallahassee at the house of the Florida insurance commissioner. I said, 'One thing you have to be aware of is that your (insurance) industry is taking quite a hit. And you better send teams of examiners to those companies as soon as possible.' He laughed and said, 'John, I'm not worried.'"
Richards was right, of course. Andrew caused about $16 billion in insured damages that triggered more than 600,000 claims. Eleven insurance companies went belly up, and the finances of dozens of others were shaken.
Tens of thousands of policyholders were left stranded. While those in Andrew's relatively narrow path began rebuilding, Allstate announced plans to cancel 300,000 of its 1.1 million policies in Florida and raise rates 32 percent, a plan it scaled back after furious protests from homeowners.
Richards said the relationship between insurance companies and homeowners began to sour.
"After Hugo, lawsuits against insurance companies were in the dozens, but after Andrew, they were in the thousands. It shows there was a different climate after Andrew."
Hurricane Andrew was particularly helpful, though, to a woman in Boston named Karen Clark, head of a little-known company called Applied Insurance Research.
While other insurers had predicted potential insurance losses in the low billions of dollars, Clark had warned that a major South Florida storm could generate a $30 billion insurance industry hit.
Her calculations were based on a novel way of looking at risk.
In the past, insurance companies tallied potential losses in a particular area and stopped writing new policies when they felt their exposure was too high. But Clark questioned how you could truly understand risk without knowing your odds of being harmed.
So she plugged historical data on hurricane strikes into computer programs along with data about homes and buildings -- potential losses for anyone who insured these structures.
Then she ran computer simulations of what might happen in various scenarios.
Her success in predicting losses before Andrew launched the new industry of "catastrophe modeling." Other companies soon created their own models, which became known in the industry as "black boxes" because their algorithms and inputs were kept secret for competitive reasons.
These mysterious black boxes would increasingly determine what homeowners from the Gulf Coast to New England shelled out in premiums.
Higher than Mississippi?
During his career as a corporate executive, Daryl Ferguson had moved to different cities across the country, but when he finally retired on a bluff overlooking the Whale Branch River near Beaufort, he thought his property insurance seemed unusually high.
Now, with his new understanding of hurricane risks in South Carolina, he was even more convinced that his rates were out of whack.
"My insurance company, USAA, is terrific, so I did a test." He asked company officials how much it would cost to insure a newly built $400,000 home in Gulfport, Miss., versus one in Beaufort County. Gulfport had been hit hard by Katrina and Rita and is considered particularly vulnerable to hurricanes. "I couldn't believe what they told me." His hypothetical Gulfport bill would be one-third the price of his Beaufort premium. "I turned to my wife and said, 'This is like Sherlock Holmes; one question leads to another.'?"
Ferguson had stumbled onto something. The average premium for $400,000 homes in South Carolina was the seventh highest in the nation, roughly the same as tornado-prone Oklahoma, according to the National Association of Insurance Commissioners.
"I was shocked," Ferguson said. "My first instinct was to go to the Department of Insurance and ask them why they're approving these rates, but I decided to do some homework first."
He began looking for Martin Simons.
'Impact every citizen'
Few people outside the insurance industry knew more about the black boxes than Simons, a bookish man with a scraggly salt-and-pepper beard who, unknown to homeowners, had long had a major impact on what they paid in insurance.
Between 1985 and 1997, Simons had been deputy director and chief actuary for the state Department of Insurance. In that capacity he had forced some insurers to reduce rates when their profits were too high and urged others to raise rates when his analyses showed they were too low. He left the agency when he felt state leaders had become too anti-regulation, and he had gone on to build a national reputation in the arcane world of actuarial analyses.
When Florida regulators established the nation's first independent panel to review how computer models affect rates, they hired Simons. He later did work for Maryland and California, and conveniently lived in Columbia. Ferguson met Simons for lunch at the Chili's restaurant next to Simons' granddaughter's auto repair shop in Summerville. "The first words he said to me were, 'Daryl, I know what you want.'" Ferguson was puzzled. "Then he said, 'You want to know why the Department of Insurance doesn't regulate its homeowner industry?' It was exactly what I wanted to know."
Simons had long lamented the lack of transparency about the catastrophe models. In 2003, State Farm requested a 29 percent increase in its rates. This hike would ripple across the state because State Farm controls about 25 percent of the homeowner's insurance market.
State Farm justified the increase in part because of results from the black boxes.
At the time, the state Department of Consumer Affairs was charged with reviewing rate increases, and the agency hired Simons to scrutinize State Farm's proposal. In sworn testimony before the hearing, Simons said the request was "seriously flawed," largely in part because state regulators knew nothing about how catastrophe models work.
The stakes were huge, he testified: Unscrupulous insurers theoretically could choose a model to set rates as high as possible. Or they could use a model's calculations to justify reduced rates to undercut their competitors, putting the company at risk if a storm struck.
"Eventually all property insurance premiums for hurricane coverage in South Carolina will be determined using the outputs of stochastic computer hurricane simulation models," he testified, adding that the insurance industry also uses these black boxes to assess terrorism risks and in health, auto and life insurance calculations.
How the state deals with these models "will impact every citizen in this state."
Minutes before the rate case was set to begin, State Farm settled for a 19 percent increase and an agreement that it wouldn't oppose an independent panel to examine catastrophe models.
That review didn't happen.
The Department of Insurance, with money from the S.C. Sea Grant Consortium, asked three experts to look at the models. One was Peter Sparks, a noted civil engineering professor at Clemson University who had done extensive studies about wind speeds and damage.
Sparks has found that the National Hurricane Center tends to overstate wind speeds inland, and that "an unwise modeler using National Hurricane Center reports could easily get a distorted picture of the wind climate and recommend rates far higher than justified."
He said that when he and the other panel members asked the modeling companies for a look at the assumptions built into their black boxes, "They said that the information was proprietary and would not disclose their methods. We said we could not judge the soundness of them, and the Department of Insurance eventually abandoned the whole exercise."
Meanwhile, the General Assembly eviscerated the budget of the S.C. Department of Consumer Affairs, the government's main insurance watchdog.
The department lost half of its employees over the past five years; today, it has about 30 staff members to cover thousands of consumer complaints and insurance issues, one-third the roster of the University of South Carolina football team.
In 2007, amid threats that insurance companies would abandon coastal areas, lawmakers also stripped the agency of its ability to challenge rate increases below 7 percent. That meant, in effect, that an insurance company could raise rates almost at will as long as its average increase was below 7 percent. "We've had a lot of 6.9 percent rate increases since then," said Elliott Elam, the state's consumer advocate.
To Simons, insurance rates involve finding a delicate balance between a company's capacity to make money and a homeowner's ability to pay. And in the absence of a serious review of these black boxes, he feared that homeowners on the coast are getting crushed.
Ferguson said alarm bells went off when he heard Simons talk about this imbalance.
As a former CEO of large and heavily regulated utilities, "I had first-hand knowledge about what happens when a state doesn't regulate," he said. It meant millions of dollars could be added to the company's bottom line, and millions subtracted from customers' pockets.
Not a crystal ball
Experts acknowledge that no computer model can predict the future, but a new iteration of computer models has taken a step in that direction. These new models use data on warming oceans and other variables to forecast potential hurricane losses in a five-year period. In the mid-2000s, these models predicted a major increase in hurricane losses, and as a result, insurance companies sought rate increases and pulled out of coastal areas in South Carolina and elsewhere, triggering what was widely described as an "insurance crisis."
So far, the predictions of these new models haven't panned out, said Karen Clark, the architect of the original black box. In a study, she found that they overestimated losses by as much as $53 billion.
Clark told The Post and Courier that catastrophe models are useful but crude tools, and that it makes little sense to predict hurricane losses in the near term when meteorologists struggle to predict how many storms might form in the coming hurricane season.
"Trying to project hurricane experience over a short-term horizon is inherently flawed," she said.
Then again, the use of catastrophe models has "added a degree of stability to the insurance market that wasn't there before them," said Michael Young, a senior director with Risk Management Solutions, the largest of the modeling companies. The industry's performance is proof, he and other insurance experts said.
While many companies went belly-up after Hurricane Andrew, only one went under after the devastating hurricanes in 2004 and 2005. Last year, despite horrific losses worldwide and in the United States, the U.S. property and casualty industry made $22 billion in profits.
In Beaufort, Ferguson pages through notebooks he has compiled during his investigation. From his upstairs window, the marsh in the Whale Branch River shimmers in the heat. He estimates that he has spent 2,000 hours on this project, time that gave him new appreciation of the mysteries of the Lowcountry and fewer reasons to fear that a hurricane will blow it all away.
"We have plenty of time to get out of the way if a storm does come our way, so the risk isn't about people anymore; it's about property." As he grew less fearful about hurricane season, he thought about the opportunities this new perspective presents: How a regional campaign to promote Lowcountry tourism in the fall could generate thousands of new jobs, how an in-depth look at insurance rates by state regulators could stimulate the economy like a tax cut.
With homeowners shelling out $1.3 billion in premiums, even a small percentage reduction could mean tens of millions of dollars.
But as stores put up hurricane displays, and emergency officials issue fresh warnings to get ready for the hurricane season, the questions in Ferguson's mind continue to spin, especially when he goes to a picnic on Memorial Day and hears what happened to his friends. They live in Bull Point, a subdivision well inland from Beaufort, and they had just received a letter from their insurance company canceling their insurance effective Aug. 29. The reason: catastrophic wind exposure.
"Why?" he asked out loud after seeing the letter. The company hadn't explained its reasoning, or given his friends a chance to discuss the issue. "Overall, this looks like a monopoly gone wild," he said, "and the only one that loses is the customer."